The Business School Was Built For Another Technology

There is a famous story about why American railroad tracks are 4 feet, 8½ inches wide.

The story goes that Roman chariots helped determine the width of Roman roads. Those roads influenced the width of wagons, wagons influenced early British railways, and British railway standards crossed the Atlantic. Centuries later, American trains were still running on dimensions supposedly inherited from the Roman Empire.

Historians debate the details. But the story illustrates a powerful principle: a standard can outlive the reason it was created and eventually become a constraint nobody questions.

Few engineers today would argue that 4 feet, 8½ inches is the scientifically optimal width for railroad tracks. It is simply the width the system inherited. Changing it would require redesigning an enormous infrastructure built around it.

Business schools may be facing exactly the same problem with the 90-minute class.

Few educators would argue that 90 minutes is the scientifically optimal duration for learning. Yet business schools build programs, faculty workloads, student schedules, and classrooms around it.

Why?

Because the 90-minute class made sense for the written case method.

Students prepared individually, came to class, and faculty orchestrated a discussion in which competing interpretations were debated, assumptions challenged, and managerial judgment developed collectively. When faculty were the primary gateway to expertise, concentrating learning into scheduled classroom sessions was a sensible response to the scarcity of faculty attention.

But the underlying technology has changed.

Generative AI gives students continuous access to explanations, feedback, alternative perspectives, simulations, and opportunities for practice. Whether schools encourage it or not, students are already using AI before, between, and after classes.

Yet much of the institutional conversation starts with a different question:

“How do we use AI in a 90-minute class?”

That may be the wrong question.

It assumes the 90-minute class is fixed and that AI simply needs to be fitted inside it. AI cases, AI role plays, AI tutors, and AI simulations are all squeezed into an architecture designed for something else.

The more important question is:

If we were designing this learning experience from scratch today, with AI available from day one, would we still design it around a 90-minute class?

Sometimes, absolutely.

A difficult strategic debate, a live negotiation, or a discussion that depends on many perspectives colliding in real time may need 90 minutes or more.

But other learning experiences might call for something entirely different. A short faculty intervention followed by individual exploration. An intensive simulation followed by reflection. Several shorter encounters rather than one long session. A sequence of experiences distributed across a week.

The point is not to eliminate the 90-minute class.

It is to stop starting there.

The timetable should be an outcome of learning design, not its starting point.

LOOK AT TELEVISION

Television faced a remarkably similar problem.

For decades, television was organized around a fixed unit: one episode, released at the same hour, once a week. That schedule was not dictated by storytelling. It was dictated by technology, limited channels, advertising, and the economics of broadcasting.

When digital distribution removed those constraints, much of the industry kept the old structure anyway.

Netflix did something different.

In 2013, it released the entire first season of House of Cards at once, challenging the assumption that great television had to be delivered one episode per week.

The content had not changed.

The unit of delivery had.

Business schools now face a similar choice.

The 90-minute class was never a law of pedagogy. It was a sensible response to a world in which faculty expertise was scarce and largely had to be delivered synchronously, in one room, at one time.

AI changes that constraint.

The question is therefore not how to make AI fit the existing timetable.

It is what unit of teaching each learning experience actually requires.

THIS IS A LEADERSHIP ISSUE

For deans, this is more than a teaching innovation. It is a governance and resource allocation question.

Faculty workload models often reward scheduled contact hours. Timetabling systems assume fixed blocks. Classroom investments reinforce the same architecture. Course structures and programs calendars make changing it difficult.

But those systems were designed around a world of scarce faculty attention.

If the technology of learning has changed, the infrastructure supporting it may need to change as well.

That does not mean abandoning classrooms, fixed schedules, or the case method. It means giving faculty and program designers the freedom to ask a more fundamental question:

What is the best learning experience we can create, and what timetable does that experience require?

That may lead to more classroom time in some places, less in others, shorter sessions, longer sessions, intensive simulations, asynchronous work, or unexpected interventions when a real-world event makes a concept suddenly relevant.

The key is that the learning design should determine the timetable, not the timetable determine the learning design.

And the implications extend well beyond the 90-minute class.

The semester itself is an inherited unit. Why should every course still be divided into twelve or fourteen weekly sessions simply because that is how the academic calendar evolved? Some learning may require concentrated bursts. Some may benefit from shorter, more frequent interactions. Some may unfold over several weeks as students make decisions and experience their consequences.

The same question applies to faculty workload, program architecture, classrooms, and even future capital investments.

If teaching is no longer confined to scheduled blocks, workload models based primarily on contact hours will measure the wrong thing. If learning no longer has to happen in fixed blocks, timetabling systems and physical infrastructure designed around those blocks can become constraints rather than enablers. And if the architecture is changing, building more capacity around the old architecture may simply lock schools further into it.

This is why the issue cannot be left to individual faculty members or isolated AI pilots. No professor can redesign the timetable, workload model, semester structure, program architecture, and physical infrastructure alone.

This is a leadership decision.

Business schools teach their students to question the assumptions embedded in an industry’s existing architecture. They teach them to ask whether rules that once made sense still make sense when technology and markets change.

AI is now asking business schools to do the same to themselves.

This is no longer only a pedagogical question. It is a strategic one.

And it will not be decided by any single class or course. It will be decided by whether business school leadership is willing to do to itself what it teaches its own students to do to their industries: look past the current architecture, question its inherited assumptions, and design for the world that already exists rather than the one that used to.

The classroom has lost its monopoly on learning.

Perhaps, for the first time in a century, the timetable has lost its monopoly on teaching.

The question is whether business schools will redesign around what learning now makes possible or simply fit a new technology onto an old track.

---

Ithai Stern is a Professor of Strategy at INSEAD. Benjamin Stevenin is the former Director of Business School Solutions and Partnerships at Times Higher Education.

© Copyright 2026 Poets & Quants. All rights reserved. This article may not be republished, rewritten or otherwise distributed without written permission. To reprint or license this article or any content from Poets & Quants, please submit your request HERE.